Can Ubuntu Linux ever pay for itself? The conventional wisdom is that it can't, because no distribution has done so in the past. However, that doesn't stop Canonical, Ubuntu's commercial arm, from trying hard. At the very least, Canonical is trying to defray as much of the cost as possible.
Canonical is not a publicly traded company and does not release any financial figures. The company is quick to announce distribution deals, but the value of those deals are noticeably absent from many of its news releases. Ask its public relations directly for such information, and you are told that it is "confidential." Nor is this lack of information surprising, since, from a traditional business perspective, Canonical has nothing to gain from transparency.
Under these conditions, all answers to such questions must remain speculative. Furthermore, the answers can vary widely, depending on the assumptions made.
All the same, some guesses can be pieced together that have a good chance of being somewhere on the playing field, even if they are not in contact with the ball.
History suggests that making Ubuntu self-sustaining is impossible. Traditionally, it is services and support contracts that have supported companies that produce distributions. The history of Linux is littered with distributions that have tried to pay for themselves, and every single one of them has failed. They include Corel, Linspire, Mandrake, Progeny, Stormix, and SuSE -- and those are only the first that come to mind.
Not that Ubuntu is strictly comparable to any of these commercial predecessors. For one thing, it is more polished than any of them, although some were considered polished for their time.
For another, unlike many of its predecessors, Ubuntu has never tried to sell a box packaged for stores. With all these examples of how not to, Ubuntu has generally avoided such obvious mistakes.
Instead, Canonical has tried to emulate -- at least intermittently -- successful distributions like Red Hat, whose success is built on services rather than software. From the variety of Canonical's announcements, it is still trying to find the combination of offerings that will lead to success, but its circumstances are not strictly comparable to any of these previous distributions.
All the same, Canonical never seems to have completely abandoned the idea of making at least some money from the desktop. From time to time, it announces pre-load deals, and its desktop is becoming crowded with services intended to appeal to the desktop user. Those services include Ubuntu One's cloud services, its music store, and the controversial addition of Amazon search results. Similarly, the Ubuntu download page recently added a donation page that must be clicked through before the download begins.
The income from all these desktop services is even more unknown that the total cost of making a distribution. Some of these services are too new to have any track record, and Canonical releases results for none of them.
However, the fact that others keep being added suggests that Canonical has yet to break even. My own admittedly unsupported guess is that a large percentage users have little interest in such services, if only because those who might use them are already using other services.
For instance, if you are already using iTunes, what can the Ubuntu music store offer that you don't already have? Not greater selection or consistently lower prices. It may be better integrated into the desktop, but if users have become accustomed to accessing iTunes through their Web browser, they have probably stopped noticing any inconvenience long ago.
Another reason for thinking that income from the desktop remains small is that Ubuntu has not always looked for new services wisely. Only last year, headlines were made when Ubuntu tried to take all the affiliate revenue derived from users of the Banshee music player buying from online stores, rather than leaving it for Banshee's developers. This cash grab could have simply been evidence of inexperience, but it might also suggest a certain desperation.
Some of these suggestions are probably erroneous. All the same, once you start thinking, you can hardly avoid the conclusion that releasing two versions of a distribution every year results in tremendous overhead for Canonical, and that extraordinary efforts are needed to reduce that overhead to any degree.